Policy

What it is

In practice, a policy is a function or model that accepts an input state and returns a probability distribution over possible actions. If you are using a transformer for agentic workflows, the policy is effectively the weights of the model that decide the next token to emit. These models are often trained via reinforcement learning where the policy is iteratively updated to maximize a cumulative numerical reward. The size of the policy is equivalent to the parameter count of the underlying model, ranging from millions to hundreds of billions of parameters.

Why it matters

Understanding policy is critical when you deploy agents that interact with external APIs or users because the policy determines the reliability and safety of those interactions. If you do not monitor the policy behavior, an agent may fall into action loops or take unauthorized actions. You should care because tuning the policy, often through fine-tuning or prompt engineering, is the primary way to reduce error rates in autonomous workflows. Ignoring the policy risks unexpected runtime behaviors that can lead to high latency or excessive API costs due to suboptimal decision-making.

In practice

In production, you influence the policy by updating the system prompt, providing specific in-context examples, or swapping to a fine-tuned model version. You observe policy behavior by tracking action success rates or using logging to audit the agent's decision paths against your intended logic. When the policy fails, you generally adjust the training data or the system instructions rather than modifying the inference engine directly.

The tradeoff

The primary tradeoff is between a policy that is too rigid, causing failure on edge cases, and one that is too flexible, resulting in unpredictable or hallucinated actions.

Where it appears

Research summaries that use Policy, each linked to its source paper.

• Fixing Stale Spatial Memory in AI Agents
  When Memory Lies: An Empirical Study of Spatial Memory Staleness in VLM Agents

  The paper introduces a read-time filtering intervention called OMCD to prevent vision-language model agents from making fatal navigation errors caused by outdated spatial knowledge.

• Evolving Rubrics for Better Audio Reasoning
  Reinforcement Learning with Evolving Rubrics as Rewards for Audio Reasoning

  The researchers improved audio-language model reasoning by replacing static rewards with dynamically evolving, audio-grounded rubrics during reinforcement learning.

• Enabling Native Computer Use for Agents
  Qwen-CUA: Native Computer Use for (almost) Everything

  The paper introduces Qwen-CUA, a system that enables autonomous agents to operate desktop and web software natively by processing visual screenshots instead of relying on specialized APIs or accessibility metadata.

• Why Robots Work Better in Chunks
  Why Does Action Chunking Improve Behavioral Cloning Performance in Robotic Control?

  The paper examines why grouping robotic actions together improves control performance, identifying that the benefits come from better error management and learning multiple ways to complete tasks.

• Building 3D Designs One Step at a Time
  CADENA: Stepwise CAD Reverse Engineering

  The paper introduces CADENA, a system that reconstructs 3D computer aided design models by building them sequentially while checking for geometry errors at each step.

• Direct Preference Optimization for Language Models
  Direct Preference Optimization: Your Language Model is Secretly a Reward Model

  The paper introduces Direct Preference Optimization as a simpler method to align language models with human preferences without using reinforcement learning.